MilSim
Building a wargame to explore AI battlefield prediction
About this project
The accepted First Playable MVP v1 provides a deterministic text-mode command loop with delayed reports, role-limited perception, JSON scenarios, scoring, debrief, and objective replay. A local browser client with an interactive tactical map, route planning, and scenario controls is implemented but remains in review.
Updated on
17.06.2026
Version
v1.0.0
Project brief
Agent-ready context
Problem
Most strategy systems give the player direct access to objective state, while real command decisions depend on delayed, incomplete, role-limited, and increasingly stale information.
System
A headless-first C# and .NET simulation separates objective battlefield state from a commander's perceived state through deterministic events, delayed reports, role views, and replaceable CLI and browser clients.
Input
JSON scenarios, terrain, unit state, command authority, player orders, simulation ticks, detection events, and delayed reports.
Output
A playable command loop with role-limited tactical views, route planning, contact reports, scoring, debrief, and objective replay.
Evidence
The clean committed baseline was freshly verified on 2026-09-13 with 143 of 143 tests passing, a successful web build, JavaScript syntax checks, and an HTTP smoke test covering three scenarios and a 400-cell map.
Status
Accepted text-mode First Playable MVP v1 with a local browser prototype in review. Enemy AI, multiplayer, save and load, scenario editing, mature combat, packaging, and public deployment remain outside the verified release.
